Can Double Auctions Control Monopoly and Monopsony Power in Emissions Trading Markets?

Can Double Auctions Control Monopoly and
Monopsony Power in Emissions Trading
Markets?
R. Andrew Muller
McMaster University, Hamilton, Ontario, Canada
Stuart Mestelman
McMaster University, Hamilton, Ontario, Canada
John Spraggon
Lakehead University, Thunder Bay, Ontario, Canada
Rob Godby
University of Wyoming, Laramie, Wyoming, U.S.A.
This Revision: September 28, 2000
Corresponding Author: R. Andrew Muller
Department of Economics
McMaster University
Hamilton, Ontario
Canada L8S 4M4
mullera@mcmaster.ca
Proposed Running Head:
Market Power and Emissions Trading
JEL Categories:
Q28, C91, D42

---

Can Double Auctions Control Monopoly and Monopsony Power
in Emissions Trading Markets?
Abstract
We conduct a laboratory experiment to investigate whether the double auction institution can suppress
market power in emissions trading markets. We study twenty-four markets with varying market structure
in a ABA crossover design which controls for subject effects. We find clear evidence of successful use of
market power. Average prices rise under monopoly and fall under monopsony. Opening prices are
affected much more than closing prices. Profits are redistributed in favour of the agent with power.
Efficiency is not affected significantly. Analysis of convergence trends suggests this is not a transitory
phenomenon. We interpret our results as evidence of successful price discrimination within a double
auction market.
Key Words: Emissions trading, double auctions, market power, tradable permits, price discrimination
JEL Categories: Q28, C91, D42

---

Table of Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Experimental Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Benchmarks and Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Average Prices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Coupon Sales and Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Speculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Profits, and Effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Price Discrimination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Price Convergence Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

---

Can Double Auctions Control Monopoly and Monopsony Power in
Emissions Trading Markets?1
Introduction
Emissions trading is frequently advocated as an instrument for market-based environmental
regulation. Unfortunately many potential emissions trading markets are likely to be sufficiently
concentrated to create market power. At the international level, it is frequently thought that the United
States will be a dominant buyer and the states of the former Soviet Union dominant suppliers in Annex I
trading under the Kyoto protocol. Nordhaus and Boyer (1999, 121) estimate that the United States will
account for approximately 44 percent of carbon emissions permit purchases by 2010 while the former
Soviet Union will account for nearly 56 percent of sales. By 2050 the share for the United States will fall
to 39 percent while the share for the former Soviet Union will rise to 68 percent (in both 2010 and 2050
eastern Europe will account for the balance of sales). These estimates are based on competitive market
pricing. Bernstein et al. (1999, 250) estimate that full exploitation of monopoly power by Eastern Europe
and the former Soviet Union could induce a monopoly mark-up of 180 percent and raise international
carbon permit prices from US$90/tonne to $129 per tonne.
These concentration concerns are particularly relevant if trading under the Kyoto Protocol is
implemented on a country-to-country basis. If countries delegate trading authority to polluting firms,
concentration in world greenhouse gas markets could be significantly lessened. High market concentration
may still be a problem in trading permits for other pollutants, however, particularly in markets for
regionally restricted air and water pollutants. For example, the Ontario Ministry of Environment (2000)
has announced a mandatory cap on nitrogen oxide (NO ) and sulphur oxide (SO ) emissions from six
x
x
1 The funding for the laboratory sessions described in this paper was provided by a McMaster
University Arts Research Board grant to Muller. The paper has benefited from comments by Tim Cason,
Dan Friedman, Rob Moir, seminar participants at a number of universities and several anonymous referees.
1

---

2
fossil fuel generating stations in the beginning of 2001. The generating stations are all owned by Ontario
Power Generation (OPG), but these may be sold in the future. Current production of NO is about 50
x
kilotonnes (kt) per year, well in excess of the cap of 36 kt. The shortfall can be made up by purchases of
emission reduction credits from sources in the non-capped sector. The market for these credits is
geographically restricted by a requirement that they be generated by sources in or (to a limited extent)
upwind from Ontario. Specific measures are proposed for discounting credits generated more than 300 km
upwind of the region. OPG has undertaken an extensive program of amassing and banking these credits.
There are no other purchasers. Off-the-record comments from industry observers suggest that OPG is
paying distinctly less for these credits than might be expected in the United States. This is consistent with
the exercise of monopsony power.
The effective exercise of market power in emissions trading markets might raise concern on two
distinct grounds. First, of course, market power may restrict the net sales of permits and lead to an
inefficient allocation of responsibilities for abatement. Secondly, market power may reallocate the
allocation of gains from trade in a direction that may or may not be viewed favourably, depending on one’s
political perspective. Note that it is possible to have the second effect without the first if the traders with
market power are able to practice price discrimination. Porter (1991) has conjectured this is a possibility
in double-auction markets.
Some observers have suggested that potential market power in emissions trading may be controlled
by using double auctions (Bohm 1998, 55).2 This suggestion is based on laboratory evidence. Smith
(1981) found that monopolists trading in a double auction market experienced difficulty in maintaining
monopoly prices. Monopolists in his double oral auctions were able to obtain only about 25 percent of the
2 A double auction is organized like most stock markets. Any agent may submit a bid to buy, an
offer to sell, or an acceptance of an outstanding bid or offer. Outstanding bids, offers, and transactions
prices are common knowledge.

---

3
potential monopoly price increase.3 They achieved 100 percent of the potential increase under a posted-
offer institution.4 Smith and Williams (1989) replicated this experiment and found much lower prices; in
the last periods of their sessions, monopolists achieved only about 6 percent of the potential price increase.
Smith and Smith and Williams explain these results by postulating that buyers’ resistance to high prices
(which they term “tacit collusion”) is increased once the monopolist reveals himself prepared to sell at
lower prices.
Further experiments have confirmed that firms can exercise market power more easily in a posted-
price environment than in a double auction environment. After summarizing laboratory work in this area,
Holt (1995, 398) concludes that “sellers are sometimes able to exercise market power in double auctions,
but the influence of seller market power is much weaker (in the double auction) because of the incentives to
offer last-minute price concessions and the more active role that buyers have in this institution.” Since the
monopolist cannot pre-commit to the monopoly price as trading progresses, it lowers its price when gains
from trade still exist at the end of a trading period. As these price reductions become public knowledge, in
succeeding periods more buyers wait for the price reductions, causing the monopolist’s market power to be
eroded further, as it may be more profitable for the monopolist to make some sales at competitive prices
than no sales at all.
Although existing evidence demonstrates that double auctions are relatively successful at
controlling market power, we argue that this evidence is too weak to presume that double auctions are
sufficient to discount concerns about concentration in emissions markets. First, the claims for the double
3 The potential price increase is derived from predictions for a monopolist who posts a single price
throughout a period. The double-auction data are based on the last period in each of three sessions of
differing length. Prices were declining during the last few periods of each session, suggesting that they
might ultimately have converged to an even lower value. This decline was not evident in the posted offer
session.
4 In a posted-offer market sellers announce prices which remain fixed throughout the trading
period.

---

4
auction are based on very few laboratory sessions (three for Smith, five for Smith and Williams).
Second, Smith’s original experiments showed substantial output restriction under monopoly (on average his
monopolists traded 2.28 units less than the competitive output, or 76 percent of the predicted output
restriction of three units), so that efficiency may suffer in double auction markets even if prices are not
raised. Third, these experiments did not test whether observed prices were converging to competitive
levels nor did they provide a controlled contrast between competitive and monopoly environments.
It may be particularly difficult for double auctions to control market power in emission trading
markets. These markets exhibit some special features not always found in the paradigmatic buyer/seller
markets investigated by Smith. Smith’s sellers are given marginal cost schedules; they earn no profits at all
unless they trade. In contrast, sellers of emissions permits generally have the option of earning profits by
using unsold permits in their own operations. Selling a permit may have a clearer opportunity cost and this
may create a frame in which sellers may be less vulnerable to counter-withholding by buyers. Another
special feature of many emissions markets is that agents can act as traders, buying and selling permits for
resale or repurchase. This does not affect predicted prices under competition or single price monopoly, but
it allows speculation and consequently may introduce more noise into the price structure and lead to
unknown dynamic effects.
Several experiments have detected market power being exercised by dominant firms in double
auction markets for tradable emission permits. Ledyard and Szakaly-Moore (1994) adapted Smith’s
parameters to an emissions trading environment by allowing all agents to buy and sell coupons. In two of
three sessions they discovered a “strong” monopolist who was able to achieve earnings close to those
predicted for a single price monopolist. Even in the last period of the experiment the strong monopolists
were able to achieve over 100 percent of the potential monopoly price increase (the weak monopolist

---

5
realized approximately 16% of the potential monopoly price increase).5
Brown-Kruse, Elliott, and Godby (1995) also detected market power in laboratory markets related
to emissions trading. In their experiment single buyers or single sellers of emission permits with a capacity
to use ten permits participated in a market with ten sellers or ten buyers of one permit each. The value of a
permit was derived from the cost savings it permitted the subject. The subject with market power had
information about the cost schedules of the remaining participants but could not identify which trader
owned which schedule. In the last period of their sessions, monopolists achieved an average of 40 percent
of the potential price gain and monopsonists 166 percent of the potential price reduction.6
Godby (1997, 1999, 2000) replicated many of the results from Brown-Kruse et al. in a market
that allowed trading (i.e. buying and selling by the same agent). He aggregated the ten smaller agents in the
Brown-Kruse et al. experiment into five composite subjects with the capacity to use two permits each,
while retaining a single buyer or single seller with the capacity to use ten permits. All subjects were
permitted to both buy and sell permits in the same period. In the last period of the experiment his
monopsonists achieved an average of 147 percent of the available price reduction. His monopolists were
not so successful. Despite a pattern of high prices in earlier periods, by the last period final transaction
prices in the monopoly market were below competitive levels.7
5 This summary is based on the graphical presentation of their MUDA sessions (their figures 6
and 7).
6 These data refer to simple manipulation. Brown-Kruse et al. also considered strategic
manipulation, under which agents with market power had an incentive to alter their permit purchases to
affect conditions in a downstream market.
7 Only two of the three sessions exhibited this outcome and then only in the last period of the
monopoly treatments. Prices in the final periods of these sessions were lower but much more volatile than
in previous periods. It is not clear whether this was a transitory phenomenon or an indication that these
markets were beginning to converge to a competitive price outcome after a number of stable monopoly
pricing outcomes.

---

6
In short, the limited laboratory evidence suggests that it is premature to be sanguine about the
ability of double auctions to control market power in emissions trading markets. Further investigation is
required before reaching a definite conclusion. In particular, it would be useful to directly contrast the
performance of competitive and monopoly markets under the same set of underlying cost conditions. This
paper presents such an experiment. We create a laboratory market in which ten traders are given
redemption value schedules and allowed to trade permits. Five are expected to be net buyers of permits,
five net sellers. We then aggregate either the five buyers into a single monopsonist or the five sellers into a
single monopolist. Thus we consider market power both on the selling side (monopoly) and on the buying
side (monopsony). Unlike all previous experimenters, who have adopted a between-sessions design for
investigating this problem, we adopt an ABA crossover design obtain control over subject effects by
allowing the same group to participate in both market power and competitive environments. Within any
one session we switch between competition and a market power and back again. This allows a powerful
contrast between the two market structures.
Experimental Design
We created a market environment in which ten subjects traded ration coupons in a computerized
double auction. Subjects were informed that they each represented a firm which produced a product from
several inputs.8 Each period the firm received revenues from selling its product and incurred costs from
purchasing inputs. One input, called leets, was rationed and could only be obtained by surrendering a
ration coupon. No additional payment was required. Using leets increased profits by reducing other
operating costs. Thus the marginal value of a ration coupon was equal to the increase in total operating
profit induced by employing one more unit of leets. Coupons were distributed at the beginning of each
period to some subjects but not to others. Subjects receiving coupons could use them to increase operating
8 Visit www.socsci.mcmaster.ca/~econ/mceel/abstracts/etmp-instr.htm for the complete
instructions.

---

7
profits or sell them to other subjects. Subjects who did not receive coupons could choose to buy them.
Once bought, these coupons could be used to increase operating profits or resold to other subjects. Thus
all subjects were allowed both to buy and sell coupons, but no short sales were permitted. Coupons could
not be carried over to future periods.
Clearly, the use of leets is analogous to the use of the environment to assimilate emissions and
ration coupons are analogous to annual emission permits. We adopted the leets/coupons terminology in
order to prevent the subjects’ being influenced by emotional reactions to the concept of emissions trading
and to be consistent with the terms used in the software.
Subjects traded coupons in a computerized double auction market.9 They were guided in their
trading by a wizard, a small window on their computer screens which informed them how much adding or
subtracting one coupon from their holdings would change their operating profits. A market consisted of a
number of trading periods. At the end of each trading period, subjects were informed of the redemption
value of their coupon holdings, their net sales revenue and their operating profit for the period. Total
earnings, including profits from trading, were continuously displayed in an inventory screen. Cumulative
earnings for the market were displayed after the last period in each market.
Each experimental session contained four markets: one practice market lasting two or three periods
of 10 minutes each and three regular markets lasting for 10 periods of three minutes each. During the
unpaid practice market subjects were carefully instructed in the use of the software. Following this,
subjects participated in the three “real” markets for that session. Subjects recorded their earnings at the
end of each regular market. At the end of the session they were paid their earnings privately in cash.
The redemption values used in the experiment are derived from those used by Smith (1981), Smith
9 We used the RNSC double auction reported in Godby, Mestelman, Muller and Welland (1997)
and described in more detail in Mestelman and Muller (1998).

---